Bimetallic Reductive Elimination from Dinuclear Pd(III) Complexeswag.caltech.edu/publications/sup/pdf/886.pdf · 2010-11-02 · Bimetallic Reductive Elimination from Dinuclear Pd(III)

Bimetallic Reductive Elimination from Dinuclear Pd(III)Complexes

David C. Powers,† Diego Benitez,‡ Ekaterina Tkatchouk,‡ William A. Goddard III,‡

and Tobias Ritter*,†

Department of Chemistry and Chemical Biology, HarVard UniVersity, 12 Oxford Street,Cambridge, Massachusetts 02138, and Materials and Process Simulation Center, California

Institute of Technology, Pasadena, California 91125

Received April 29, 2010; E-mail: [email protected]

Abstract: In 2009, we reported C-halogen reductive elimination reactions from dinuclear Pd(III) complexesand implicated dinuclear intermediates in Pd(OAc)2-catalyzed C-H oxidation chemistry. Herein, we reportresults of a thorough experimental and theoretical investigation of the mechanism of reductive eliminationfrom such dinuclear Pd(III) complexes, which establish the role of each metal during reductive elimination.Our results implicate reductive elimination from a complex in which the dinuclear core is intact and suggestthat redox synergy between the two metals is responsible for the facile reductive elimination reactionsobserved.

Introduction

Metal-metal redox cooperation during catalysis can potentiallylower activation barriers of chemical transformations and thusallow access to reaction pathways that are difficult to accesswith mononuclear catalysts.1 Redox chemistry of multinuclearcomplexes has been intensely studied, in part due to potentialadvantages in catalysis; however, metal-metal cooperationduring redox chemistry has been difficult to establish. Mono-nuclear Pd(IV) complexes have been suggested as intermediatesin palladium-catalyzed C-H oxidation reactions since 1971.2

In 2009, we proposed dinuclear Pd(III) intermediates in catalysisas an alternative to Pd(II)/Pd(IV) redox cycles.3 In a preliminarystudy of C-Cl reductive elimination from the dinuclear Pd(III)complex 1 (eq 1), we were unable to establish whether reductiveelimination proceeded with simultaneous, co-operative redoxparticipation of both metals or via one-centered redox chemistry.Herein, we report a detailed investigation of the mechanism ofC-Cl reductive elimination from 1 and present a tool forevaluating metal-metal synergy during redox transformationsof multinuclear complexes. Our results implicate synergisticredox chemistry of both metals during reductive elimination,which leads to an energetic advantage as compared to relatedmonometallic reductive elimination reactions. This analysis isthe first detailed evaluation of reductive elimination from Pd(III)

and, more importantly, gives insight into evaluating and under-standing metal-metal redox synergy in general.

Redox catalysis in synthesis is often accomplished byhomogeneous, mononuclear catalysts.4 Oxidative addition andreductive elimination, two of the fundamental redox transforma-tions of organometallic chemistry, have been well studied formononuclear complexes. Oxidative additions are reactions inwhich the oxidation state and coordination number of a transitionmetal center increase by two upon addition of a small molecule.5

Oxidative addition can proceed via concerted,6 SN2-like,7 andradical8 mechanisms. Reductive eliminationsformally the re-verse of oxidative additionsdescribes a transformation in whichthe oxidation state and coordination number of a metal are

† Harvard University.‡ California Institute of Technology.

(1) (a) Fackler, J. P. Inorg. Chem. 2002, 41, 6959–6972. (b) Gray, T. G.;Veige, A. S.; Nocera, D. G. J. Am. Chem. Soc. 2004, 126, 9760–9768.

(2) (a) Henry, P. M. J. Org. Chem. 1971, 36, 1886–1890. (b) Stock, L. M.;Tse, K.-T.; Vorvick, L. J.; Walstrum, S. A. J. Org. Chem. 1981, 46,1757–1759. (c) Yoneyama, T.; Crabtree, R. H. J. Mol. Catal. A: Chem.1996, 108, 35–40. (d) Dick, A. R.; Kampf, J. W.; Sanford, M. S. J. Am.Chem. Soc. 2005, 127, 12790–12791. (e) Racowski, J. M.; Dick, A. R.;Sanford, M. S. J. Am. Chem. Soc. 2009, 131, 10974–10983.

(3) (a) Powers, D. C.; Ritter, T. Nat. Chem. 2009, 1, 302–309. (b) Powers,D. C.; Geibel, M. A. L.; Klein, J. E. M. N.; Ritter, T. J. Am. Chem.Soc. 2009, 131, 17050–17051.

(4) (a) Masters, C. Homogeneous Transition-Metal Catalysis; UniversityPress: London, 1981. (b) van Leeuwen, P. Homogeneous Catalysis;Kluwer Academic Publishers: Boston, 2004.

(5) (a) Heck, R. F. Organotransition Metal Chemistry: A MechanisticApproach; Academic Press: New York, 1974; p 49. (b) Miessler, G. L.;Tarr, D. A. Inorganic Chemistry, 3rd ed.; Pearson Education, Inc.:Upper Saddle River, NJ, 2004; pp 525-526. (c) Anslyn, E. V.;Dougherty, D. A. Modern Physical Organic Chemistry; UniversityScience Books: Sausalito, CA, 2006; pp 724-726.

(6) Hartwig, J. F.; Paul, F. J. Am. Chem. Soc. 1995, 117, 5373–5374.(7) (a) Chock, P. B.; Halpern, J. J. Am. Chem. Soc. 1966, 88, 3511–3514.

(b) Lau, K. S. Y.; Wong, P. K.; Stille, J. K. J. Am. Chem. Soc. 1976,98, 5832–5840. (c) Ellis, P. R.; Pearson, J. M.; Haynes, A.; Adams,H.; Bailey, N. A.; Maitlis, P. M. Organometallics 1994, 13, 3215–3226.

(8) (a) Tsou, T. T.; Kochi, J. K. J. Am. Chem. Soc. 1979, 101, 6319–6332. (b) Hall, T. L.; Lappert, M. F.; Lednor, P. W. J. Chem. Soc.,Dalton Trans. 1980, 1448–1456.

Published on Web 09/21/2010

10.1021/ja1036644 2010 American Chemical Society14092 9 J. AM. CHEM. SOC. 2010, 132, 14092–14103

reduced by 2, with concurrent formation of a small molecule.Like oxidative addition, reductive elimination can proceed viavarious mechanisms.9 In a catalytic cycle, the mechanisms ofoxidative addition and reductive elimination need not be themicroscopic reverse of one another.

Redox chemistry in biology10 and heterogeneous catalysis11

often occurs at multinuclear sites. For example, biologicallyrelevant redox catalysis, such as the reduction of dinitrogen bynitrogenase10a,b and the oxidation of methane by methanemonooxygenase,10c-e is frequently accomplished by multi-nuclear active sites. While redox catalysis involving more thanone metal is frequently encountered, it is difficult to ascertainthe specific role of each individual metal center during a redoxtransformation.

The mechanisms of redox transformations at multinuclearcomplexes are less understood than the corresponding reactionsof mononuclear complexes.12 Stoichiometric oxidative addition13

to and reductive elimination14 from dinuclear complexes havebeen observed, and dinuclear intermediates have been proposed

as intermediates in catalysis.3,15 The potential of utilizing metal-metal redox synergy to accomplish challenging transformationshas long been recognized, and while proposals of mechanismsinvolving metal-metal cooperation have been posited,16 iden-tification of the intimate role of each metal center during redoxtransformations at dinuclear complexes is difficult to establishexperimentally.14i-k,17 Understanding how metal-metal redoxsynergy can be used in catalysis requires insight into the roleof each metal during redox chemistry.

During our efforts to utilize metal-metal redox synergy incatalysis, we found it useful to employ a two-tiered nomencla-ture scheme in which organometallic redox transformations areclassified by both the nuclearity of the complex undergoing theredox transformation and the metallicity of the transformation.Nuclearity is a descriptor of structure and refers to the numberof metal centers present in the complex undergoing the redoxtransformation. Metallicity is a descriptor of the mechanism ofa redox transformation and refers to the number of metal centersthat undergo redox chemistry coupled to substrate oxidation orreduction. The presence of multiple metal centers (nuclearity)in a redox transformation does not necessitate redox participation(metallicity) of all metal centers.

• Nuclearity s descriptor of structure; the number of metalcenters in a complex that undergoes a redox transformation.

• Metallicitys descriptor of mechanism; the number of metalcenters that participate in redox chemistry during a redoxtransformation.

One can, in principle, determine the nuclearity of a redoxtransformation using the tools of reaction kinetics. For example,fragmentation of a dinuclear complex into two identical mono-nuclear complexes prior to reductive elimination can afford ahalf-order rate law with respect to the dinuclear transition metalcomplex.18 While redox transformations that take place atmononuclear complexes are by definition monometallic,19 redoxtransformations at dinuclear complexes can be either mono- orbimetallic, depending on whether one or both metals change

(9) (a) Brown, M. P.; Puddephatt, R. J.; Upton, C. E. E. J. Chem. Soc.,Dalton Trans. 1974, 2457–2465. (b) Brown, J. M.; Cooley, N. A.Chem. ReV. 1988, 88, 1031–1046. (c) Mann, G.; Baranano, D.;Hartwig, J. F.; Rheingold, A. L.; Guzei, I. A. J. Am. Chem. Soc. 1998,120, 9205–9219. (d) Widenhoefer, R. A.; Buchwald, S. L. J. Am.Chem. Soc. 1998, 120, 6504–6511. (e) Hartwig, J. F. Acc. Chem. Res.1998, 31, 852–860. (f) Williams, B. S.; Holland, A. W.; Goldberg,K. I. J. Am. Chem. Soc. 1999, 121, 252–253. (g) Williams, B. S.;Goldberg, K. I. J. Am. Chem. Soc. 2001, 123, 2576–2587.

(10) (a) Shilov, A. E. J. Mol. Catal. 1987, 41, 221–234. (b) Ericson, A.;Hedman, B.; Hodgson, K. O.; Green, J.; Dalton, H.; Bentsen, J. G.;Beer, R. H.; Lippard, S. J. J. Am. Chem. Soc. 1988, 110, 2330–2332.(c) Holm, R. H.; Kennepohl, P.; Solomon, E. I. Chem. ReV. 1996, 96,2239–2314. (d) van den Beuken, E. K.; Feringa, B. L. Tetrahedron1998, 54, 12985–13011. (e) Bioinorganic Catalysis, 2nd ed.; Reedijk,J., Bouwman, E., Eds.; Marcel Dekker, Inc.: New York, 1999. (f)Biomimetic Oxidations Catalyzed by Transition Metal Complexes;Meunier, B., Ed.; Imperial College Press: London, 2000.

(11) (a) Araki, M.; Ponec, V. J. Catal. 1976, 44, 439–448. (b) Muetterties,E. L. Science 1977, 196, 839–848. (c) Muetterties, E. L.; Rhodin, T. N.;Band, E.; Brucker, C. F.; Pretzer, W. R. Chem. ReV. 1979, 79, 91–137. (d) Shriver, D. F.; Sailor, M. J. Acc. Chem. Res. 1988, 21, 374–379. (e) Wiegand, B. C.; Friend, C. M. Chem. ReV. 1992, 92, 491–504. (f) Adams, R. D. J. Organomet. Chem. 2000, 600, 1–6.

(12) Catalysis by di-and polynuclear metal cluster complexes; Adams, R. D.,Cotton, F. A., Eds.; Wiley-VCH: New York, 1998.

(13) (a) Schmidbaur, H.; Franke, R. Inorg. Chim. Acta 1975, 13, 85–89.(b) Fackler, J. P.; Basil, J. D. Organometallics 1982, 1, 871–873. (c)Fackler, J. P.; Murray, H. H.; Basil, J. D. Organometallics 1984, 3,821–823. (d) Basil, J. D.; Murray, H. H.; Fackler, J. P.; Tocher, J.;Mazany, A. M.; Trzcinska-Bancroft, B.; Knachel, H.; Dudis, D.;Delord, T. J.; Marler, D. O. J. Am. Chem. Soc. 1985, 107, 6908–6915. (e) Elduque, A.; Aguilera, F.; Lahoz, F. J.; Lopez, J. A.; Oro,L. A.; Pinillos, M. T. Inorg. Chim. Acta 1998, 274, 15–23. (f) Jamali,S.; Nabavizadeh, S. M.; Rashidi, M. Inorg. Chem. 2005, 44, 8594–8601. (g) Abdou, H. E.; Mohamed, A. A.; Fackler, J. P. Inorg. Chem.2007, 46, 9692–9699. (h) Bonnington, K. J.; Jennings, M. C.;Puddephatt, R. J. Organometallics 2008, 27, 6521–6530.

(14) (a) Wegman, R. W.; Brown, T. L. J. Am. Chem. Soc. 1980, 102, 2494–2495. (b) Brown, M. P.; Fisher, J. R.; Hill, R. H.; Puddephatt, R. J.;Seddon, K. R. Inorg. Chem. 1981, 20, 3516–3521. (c) Hill, R. H.;Puddephatt, R. J. Inorg. Chim. Acta 1981, 54, L277–L278. (d) Hill,R. H.; de Mayo, P.; Puddephatt, R. J. Inorg. Chem. 1982, 21, 3642–3646. (e) Davies, S. G.; Hibberd, J.; Simpson, S. J.; Watts, O. J.Organomet. Chem. 1982, 238, C7–C8. (f) Sundararajan, G.; SanFilippo, J. Organometallics 1985, 4, 606–608. (g) Sundararajan, G.Organometallics 1991, 10, 1377–1382. (h) Stockland, R. A.; Anderson,G. K.; Rath, N. P. J. Am. Chem. Soc. 1999, 121, 7945–7946. (i)Stockland, R. A.; Janka, M.; Hoel, G. R.; Rath, N. P.; Anderson, G. K.Organometallics 2001, 20, 5212–5219. (j) Chaouche, N.; Fornies, J.;Fortuno, C.; Kribii, A.; Martın, A.; Karipidis, P.; Tsipis, A. C.; Tsipis,C. A. Organometallics 2004, 23, 1797–1810. (k) Ara, I.; Chaouche,N.; Fornies, J.; Fortuno, C.; Kribii, A.; Tsipis, A. C. Organometallics2006, 25, 1084–1091.

(15) (a) Misumi, Y.; Ishii, Y.; Hidai, M. J. Mol. Catal. 1993, 78, 1–8. (b)Broussard, M. E.; Juma, B.; Train, S. G.; Peng, W.-J.; Laneman, S. A.;Stanley, G. G. Science 1993, 260, 1784–1788. (c) Li, C.; Widjaja, E.;Garland, M. J. Am. Chem. Soc. 2003, 125, 5540–5548. (d) Li, C.;Widjaja, E.; Garland, M. Organometallics 2004, 23, 4131–4138. (e)Li, C.; Chen, L.; Garland, M. AdV. Synth. Catal. 2008, 350, 679–690.

(16) Nappa, M. J.; Santi, R.; Halpern, J. Organometallics 1985, 4, 34–41.(17) (a) Halpern, J. Inorg. Chim. Acta 1982, 62, 31–37. (b) Trinquier, G.;

Hoffmann, R. Organometallics 1984, 3, 370–380.(18) (a) Chan, Y. N. C.; Osborn, J. A. J. Am. Chem. Soc. 1990, 112, 9400–

9401. (b) Louie, J.; Hartwig, J. F. J. Am. Chem. Soc. 1995, 117, 11598–11599. (c) Driver, M. S.; Hartwig, J. F. J. Am. Chem. Soc. 1997, 119,8232–8245. (d) Vazquez-Serrano, L. D.; Owens, B. T.; Buriak, J. M.Inorg. Chim. Acta 2006, 359, 2786–2797. (e) Fulmer, G. R.; Muller,R. P.; Kemp, R. A.; Goldberg, K. I. J. Am. Chem. Soc. 2009, 131,1346–1347.

(19) In this conversation, we are excluding transition metal complexes inwhich significant redox participation of the ligand is involved. Fordiscussion: (a) Holm, R. H.; Balch, A. L.; Davison, A.; Maki, A. H.;Berry, T. E. J. Am. Chem. Soc. 1967, 89, 2866–2874. (b) Pierpont,C. G. Coord. Chem. ReV. 2001, 216-217, 99–125. (c) Blackmore,K. J.; Ziller, J. W.; Heyduk, A. F. Inorg. Chem. 2005, 44, 5559–5561. (d) Bart, S. C.; Chłopek, K.; Bill, E.; Bouwkamp, M. W.;Lobkovsky, E.; Neese, F.; Wieghardt, K.; Chirik, P. J. J. Am. Chem.Soc. 2006, 128, 13901–13912. (e) Haneline, M. R.; Heyduk, A. F.J. Am. Chem. Soc. 2006, 128, 8410–8411. (f) Zarkesh, R. A.; Ziller,J. W.; Heyduk, A. F. Angew. Chem., Int. Ed. 2008, 47, 4715–4718.(g) Radosevich, A. T.; Melnick, J. G.; Stoian, S. A.; Bacciu, D.; Chen,C.-H.; Foxman, B. M.; Ozerov, O. V.; Nocera, D. G. Inorg. Chem.2009, 48, 9214–9221.

J. AM. CHEM. SOC. 9 VOL. 132, NO. 40, 2010 14093

Reductive Elimination from Dinuclear Pd(III) Complexes A R T I C L E S

their oxidation state in the redox transformation. Differentiatingbetween mono- and bimetallic mechanisms is experimentallychallenging.

Mono- and bimetallic transformations are kinetically indis-tinguishable, because the transition states do not differ inchemical composition but only in specific roles of the metalcenter in the dinuclear core. Further, the identity of the transitionmetal fragment produced by a redox transformation is notsufficient to assign the metallicity of a given reaction. Forexample, the oxidative addition of MeI to dinuclear Au(I)complex 4 to afford dinuclear Au(II) complex 6 illustrates thedifficulty in assigning the metallicity of a redox transforma-tion.13d,f,h If oxidative addition proceeds initially at a single Ausite to afford the Au(I)/Au(III) mixed-valence species 5, beforecomproportionative isomerization to complex 6, the oxidativeaddition to 5 is monometallic (eq 2). Alternatively, if oxidativeaddition proceeds with Au-Au bond formation concurrent withAu-C bond formation to form 6 via 7 (eq 3), the redoxtransformation is bimetallic.

Development of catalysis concepts that rely on metal-metalsynergy during redox transformations requires fundamentalunderstanding of the role of each metal involved in the redoxtransformation. Potentially, metal-metal cooperation duringcatalysis could allow development of reactions that are difficultto accomplish with monometallic systems. We have sought toevaluate the potential advantage of utilizing two metals duringredox chemistry in desirablesbut currently challengingstransformations. Transition-metal-mediated halogenation reac-tions are difficult, and only a few methods are available.21

Reductive elimination from dinuclear Pd(III) complexes is facileand provides a venue to better understand and characterizereductive elimination from dinuclear complexes.3 Herein, weprovide evidence for synergistic redox participation of bothpalladium centers during reductive elimination and show thatbimetallic reductive elimination is more facile than relatedmonometallic processes. The observed facility of redox trans-formations at bimetallic Pd(III) complexes challenges thepreviously proposed generality of Pd(II)/Pd(IV) redox cycles.22

Results

Two fundamental questions are addressed in this manuscript.First, does C-Cl reductive elimination from 1 proceed from amono- or dinuclear complex? Second, do both metals participate

in the redox chemistry of reductive elimination? The redoxactivity of each metal is difficult to evaluate experimentally;no experimental observable can be directly correlated withreaction metallicity. We have thus pursued a computationalinvestigation of reaction metallicity. The computational modelused to assay metallicity was benchmarked by comparing thecomputed mechanism with three experimental observables: (1)the ground-state structure of 1, (2) the activation parametersfor C-Cl reductive elimination from 1, and (3) the ability ofcomputed charge distributions to correlate with experimentallyobtained F-values from Hammett analysis.

Synthesis and Structure of Pd(III) Dichloride 1. DinuclearPd(III) complex 1 was chosen as a suitable substrate toinvestigate reductive elimination from dinuclear Pd complexesfor two reasons. First, Pd(OAc)2 is frequently used as a catalystin directed C-H oxidations. Second, the bridging acetate ligandshold the two palladium centers in proximity, and therefore allowfor metal-metal interaction.23 Treatment of complex 9, derivedfrom cyclometalation of benzo[h]quinoline (8) with Pd(OAc)2,with PhICl2 resulted in the formation of dinuclear Pd(III)dichloride 1 in 92% yield (Figure 1).

By 1H NMR spectroscopy, complex 1 is diamagnetic,consistent with a Pd-Pd single bond. At -50 °C, the 1H NMRspectrum displays eight nonequivalent aromatic resonances anda single resonance for the bridging acetate ligand at 2.69 ppm.In the solid state, the Pd-Pd distance in 1 is 2.5672(5) Å, 0.27Å shorter than the corresponding distance in 9, consistent withthe formation of a Pd-Pd bond upon oxidation.24 The Pd-Obonds are nonequivalent; the Pd-O bond trans to the carbonligand is 2.133(3) Å, and the Pd-O trans to the nitrogen ligand

(20) Tolman, C. A.; Meakin, P. Z.; Lindner, D. L.; Jesson, J. P. J. Am.Chem. Soc. 1974, 96, 2762–2774.

(21) Vigalok, A. Chem. Eur. J. 2008, 14, 5102–5108.(22) (a) Kalyani, D.; Dick, A. R.; Anani, W. Q.; Sanford, M. S. Tetrahedron

2006, 62, 11483–11498. (b) Muniz, K. Angew. Chem., Int. Ed. 2009,48, 9412–9423. (c) Xu, L.-M.; Li, B.-J.; Yang, Z.; Shi, Z.-J. Chem.Soc. ReV. 2010, 39, 712–733. (d) Sehnal, P.; Taylor, R. J. K.; Fairlamb,I. J. S. Chem. ReV. 2010, 110, 824–889. (e) Lyons, T. W.; Sanford,M. S. Chem. ReV. 2010, 110, 1147–1169.

(23) Bercaw, J. E.; Durrell, A. C.; Gray, H. B.; Green, J. C.; Hazari, N.;Labinger, J. A.; Winkler, J. R. Inorg. Chem. 2010, 49, 1801–1810.

(24) (a) Cotton, F. A.; Gu, J.; Murillo, C. A.; Timmons, D. J. J. Am. Chem.Soc. 1998, 120, 13280–13281. (b) Cotton, F. A.; Koshevoy, I. O.;Lahuerta, P.; Murillo, C. A.; Sanau, M.; Ubeda, M. A.; Zhao, Q. L.J. Am. Chem. Soc. 2006, 128, 13674–13675. (c) Berry, J. F.; Bill, E.;Bothe, E.; Cotton, F. A.; Dalal, N. S.; Ibragimov, S. A.; Kaur, N.;Liu, C. Y.; Murillo, C. A.; Nellutla, S.; North, J. M.; Villagran, D.J. Am. Chem. Soc. 2007, 129, 1393–1401.

Figure 1. Synthesis of 1. ORTEP drawing of 9 and 1 with ellipsoids drawnat 50% probability (hydrogen atoms omitted for clarity). Selected bondlengths in 1 [Å]: Pd(1A)-Pd(1B), 2.5672(5); Pd(1A)-Cl(A), 2.4167(10);Pd(1A)-C(A), 2.000(4); Pd(1A)-O(1A), 2.133(3); Pd(1A)-O(2A), 2.042(3);Pd(1A)-N(A), 2.016(3). For comparison, the Pd-Pd distance in 9 is2.8419(8) Å.

14094 J. AM. CHEM. SOC. 9 VOL. 132, NO. 40, 2010

A R T I C L E S Powers et al.

is 2.042(3) Å, consistent with the stronger structural trans effectof the aryl ligand as compared to the substituted pyridineligand.25

Potential ligand fluxionality, in which the apical and bridgingligands might exchange with one another, was probed byvariable-temperature NMR spectroscopy of dinuclear Pd(III)tetraacetate 10-d6, in which the bridging acetate ligands wereperdeuterated. At -60 °C, the 1H NMR spectrum of 10-d6

displayed a single 1H NMR resonance at 1.48 ppm (apicalacetate). Upon warming of the sample to -30 °C, the integrationof the signal at 1.48 ppm (apical acetate) decreased withconcurrent development of a new resonance at 2.71 ppm(bridging acetate). Acetate scrambling was determined to befast relative to reductive elimination; warming a sample of 10-d6 to 23 °C afforded a 1:1 mixture of 12 and 12-d3 (Scheme1).26

C-Cl Reductive Elimination from 1. Upon warming ofcomplex 1 to 23 °C, 10-chlorobenzo[h]quinoline (2) was isolatedin 94% yield based on 1 (eq 1). Monitoring the thermolysis of1 by 1H NMR spectroscopy established that both decompositionof 1 and formation of 2 obey first-order rate laws. The activationparameters of C-Cl reductive elimination were determined tobe ∆H q ) 17.2 ( 2.7 kcal ·mol-1, ∆S q ) -11.2 ( 9.4 cal ·K-1,and ∆G q298 ) 20.5 ( 0.1 kcal ·mol-1 by monitoring theevolution of 2 as a function of temperature (between 5 and 35°C). The rate of reductive elimination was unaffected byexogenous chloride (up to 17.1 mM in n-Bu4NCl; 1.17 equivwith respect to 1) and acetate (up to 11.8 mM in n-Bu4NOAc;0.81 equiv with respect to 1).

Reductive elimination of 2 from 1 is accompanied by theformation of a mixture of Pd-containing products (3). Treatmentof the crude reaction mixture after reductive elimination withexcess pyridine provided a mixture of 2 and five palladium-

containing species, identified as Pd(II) complexes 13, 14, 15,16, and 17 on the basis of comparison of the 1H NMR spectrumof the mixture with the 1H NMR spectra of authentic samples(Scheme 2). By 1H NMR, the combined yield of complexes13-17 was determined to be 99%. The observed products ofreaction with pyridine establish that the oxidation state of Pdin 3 is +II. The exact structure of the Pd(II) compleximmediately after reductive elimination is unknown. (See theSupporting Information for further discussion of the identity ofthe Pd-containing byproducts of reductive elimination.)

Effect of 7-Substitution of the Benzo[h]quinolinyl Ligandon the Rate of C-Cl Reductive Elimination. Dinuclear Pd(III)dichloride complexes 18a-e with 7-substituted benzo[h]quino-linyl ligands were prepared by cyclometalation of 7-substitutedbenzo[h]quinolines with Pd(OAc)2 followed by oxidation withPhICl2. Dinuclear Pd(III) complexes with NO2 (18f) and CN(18g) substituents were not sufficiently soluble to be used inkinetics experiments. Thermolysis of 18a-e at 29 °C afforded7-substituted-10-chlorobenzo[h]quinolines 19a-e in 84-95%yield. Reaction yields did not correlate with the σ-values of thesubstituents. In all cases, C-Cl bond formation was observedexclusively; C-O reductive elimination to afford 7-substituted10-acetoxybenzo[h]quinolines was not detected. The rate ofC-Cl bond formation was measured by monitoring the forma-tion of 19a-e by 1H NMR spectroscopy. The Hammett plot27

generated from these data (Figure 2) shows that C-Cl reductiveelimination from dinuclear Pd(III) complexes is accelerated byelectron-withdrawing 7-substituents on the benzo[h]quinolinylligand (F ) 1.46).

Effect of Bridging Carboxylate Ligand on the Rate ofC-Cl Reductive Elimination. Benzoate-bridged dinuclearPd(III) complexes 20a-e were prepared by treatment of 9 with4-substituted benzoic acids at reduced pressure followed byoxidation with PhICl2. Decomposition of 20a-e at 29 °Cafforded compound 2 in 91-96% yield, with no C-O reductive

(25) Coe, B. J.; Glenwright, S. J. Coord. Chem. ReV. 2000, 203, 5–80.(26) Exchange of carboxylate ligands can proceed intermolecularly (see

Supporting Information). For simplicity, here we depict only thosecomplexes arising from intramolecular exchange; intermolecularexchange does not change our conclusions based on these experimentsbecause exchange, whether intra- or intermolecular, is fast relative toreductive elimination. (27) Hammett, L. P. Chem. ReV. 1935, 17, 125–136.

Scheme 1. Rapid Exchange between Bridging and Apical AcetateLigands Leads to a 1:1 Mixture of 12 and 12-d3 upon Thermolysisof 10-d6

Scheme 2. Treatment of the Reaction Mixture after ReductiveElimination with Excess Pyridine Affords a Mixture of 2, 13, 14, 15,16, and 17 that Establishes the +II Oxidation State for Palladium

Figure 2. Hammett plot based on 7-benzo[h]quinolinyl substitution.



elimination observed in any case. The Hammett plot27 (Figure3) generated from these data shows that reductive eliminationis accelerated by electron-withdrawing substituents on thebridging carboxylate ligands (F ) +0.71). For complexes 20f(R ) Me) and 20g (R ) OMe), bearing electron-donatingsubstituents, log(kx/kH) was not linearly correlated with thesubstituent σ-values.

Competitive Reductive Elimination of Substituted BenzoateLigands. Because the electronic properties of chloride cannotbe altered, the electronic demand of the apical ligand duringreductive elimination from 1 cannot be investigated. DinuclearPd(III) tetrabenzoates were selected as substrates to probe theelectronic demand of the apical ligand during reductive elimina-tion because the electronic properties of benzoate ligands canbe modified (Scheme 3).26 In the following experiments, the2-phenylpyridyl ligand was used in lieu of the benzo[h]quino-linyl ligand because 2-(pyridin-2-yl)phenyl benzoate is morestable toward hydrolysis than benzo[h]quinolin-10-yl benzoate,which simplified product isolation and analysis. Further, pyridineis added to the dinuclear Pd(III) complexes prior to thermolysis

because we have previously observed that, in the absence ofpyridine, the products of C-O reductive elimination can reactwith the palladium-containing byproducts of reductive elimi-nation.3b

Oxidation of Pd(II) benzoate 21 with p-nitrobenzoyl peroxide(24) afforded Pd(III) complex 22, which, after subsequentwarming to 23 °C in the presence of 20.0 equiv of pyridine,afforded a 4:1 mixture of 26 and 27, as determined by both 1HNMR spectroscopy and isolated yields of 26 and 27 (Scheme3a). Oxidation of p-nitrobenzoate-bridged complex 23 withbenzoyl peroxide (25) also afforded a 4:1 mixture of 26 and 27(Scheme 3b), which confirmed that carboxylate exchange is fastrelative to C-O reductive elimination.

Intramolecular carboxylate exchange in tetrabenzoate dipal-ladium(III) complex 22 could generate an equilibrium mixtureof three isomers (Scheme 4).26 Reductive elimination fromisomer 28 can afford either 26 or 27, while reductive eliminationfrom 22 or 29 can afford only 26 or 27, respectively. Withoutknowledge of all four product-forming rate constants, the ratioof 26 and 27 cannot be deconvoluted to provide kb/kb′, therelative rate of benzoate over 4-nitrobenzoate reductive elimina-tion.28

Oxidation of 23 with unsymmetrical peroxide 30 at -50 °C,followed by treatment with pyridine at 23 °C, afforded a 2:1mixture of 26 and 27 (Scheme 5).26 Intramolecular carboxylateexchange in dinuclear Pd(III) complex 31, which features three4-nitrobenzoate ligands and one benzoate ligand, could generateeither one of two isomers (31 or 32). Reductive eliminationfrom 32 can afford only 27 (k3), while reductive eliminationfrom 31 can afford 26 (k1) and 27 (k2). The observed 2:1 ratioof 26 and 27 shows that k1 > k2 + k3 ·Keq and that benzoateundergoes reductive elimination faster than 4-nitrobenzoate.

(28) Attempts to accomplish Hammett analysis using esp-bridged complex37, anticipated to prevent carboxylate exchange, were unsuccessfulbecause, unlike C-Cl reductive elimination from 35, C-O reductiveelimination does not proceed from esp-bridged complex 37.

Figure 3. Hammett plot based on bridging benzoate substitution.

Scheme 3. Carboxylate Scrambling on Pd(III) Complexes is FastRelative to C-O Reductive Elimination

Scheme 4. Carboxylate Scrambling on Pd(III) Complexes PreventsHammett Analysis of Electronic Demand of Apical Ligands



Reductive Elimination in the Presence of Exogenous AcOH. Wehave proposed that reductive elimination from dinuclear Pd(III)complexes similar in structure to 1 is the product-forming stepof a variety of Pd-catalyzed C-H functionalizations.3 Duringcatalysis, substrate is present in large excess relative to Pd, andthus we investigated C-Cl reductive elimination in the presenceof exogenous benzo[h]quinoline (8). By 1H NMR and UV-visspectroscopy, no interaction between complex 1 and 8 wasobserved between -50 and 23 °C. We have observed areproducible acceleration of C-Cl bond formation in thepresence of exogenous benzo[h]quinoline (8) at 23 °C.3a Wenow report a re-examination of the observed rate enhancementfor the formation of 2 from 1 upon addition of 8, which indicatesthat the observed acceleration is due to acid generated bypalladation of added 8, not N-coordination of 8 to 1, as wepreviously proposed.3a

Examination of the thermolysis of 1 in the presence ofexogenous benzo[h]quinoline (8) revealed that the concentrationof 8 decreased during the evolution of 2. (For quantitative kineticdata regarding the rate of reductive elimination from 1 as afunction of initial benzo[h]quinoline concentration, see theSupporting Information.) Cyclometalation of 8 (for example,8f9 shown in Figure 1) by the Pd(II)-containing byproductsof reductive elimination (3) generates an equivalent of acid, asa result of C-H bond cleavage during metalation. We thereforeexamined the potential effect of acid on the rate of reductiveelimination from 1 by monitoring the formation of 2 as afunction of AcOH concentration ([AcOH]). AcOH was selectedbecause potential exchange of the conjugate base, acetate, withthe acetate ligands of 1 is a degenerate process. The rate offormation of 2 was monitored at AcOH concentrations up to1.16 M in CD2Cl2 by 1H NMR spectroscopy. The rate offormation of 2 is accelerated in the presence of AcOH (Figure4). The plot of first-order rate constant vs [AcOH] has a nonzerointercept, reflective of the background rate of C-Cl reductiveelimination in the absence of AcOH, and shows that at leasttwo pathways are available for C-Cl reductive elimination: onepathway independent of and one pathway linearly dependenton the concentration of AcOH.

Computational Studies. a. Structure of 1. Description of thehalogen-metal-metal-halogen tetrad present in 1 was antici-

pated to be computationally challenging using density functionaltheory (DFT). The combination of high-oxidation-state metalcenters, metal-metal bonding, and the coupling of themetal-metal bond to the electronegative chloride ligands isdemanding for exchange-correlation hybrid density functionals.29

In addition, the lack of accurate consideration of attractivemedium-range interactions such as π-π stacking in orthodoxhybrid DFT functionals30 (e.g., B3LYP, PBE0) was expectedto overestimate Pd-Pd bond distances. The M06 functional31

has been shown to accurately describe both transition-metal-based complexes containing Au, Ru, and Pd32 and non-covalentinteractions between proximal aromatic groups.33 The M06functional, with a triple-� basis set for Pd and Cl, predicted aPd-Pd distance of 2.62 Å (compared to 2.57 Å determinedexperimentally) and accurately described the relative spacingof the benzo[h]quinolinyl ligands. (See Supporting Informationfor evaluation of other functionals.) In addition, the computedstructure of 1, which we will denote A, reflects the experimen-tally observed nonequivalence of the two oxygen-based ligandson each palladium atom (Figure 1). The M06 functional alsoaccurately described the Pd-Pd distance in 9, which does notcontain a formal Pd-Pd bond23 (experimental, 2.84 Å; com-puted, 2.87 Å). In the ensuing discussion, computed structureswill be designated by compound letters, not numbers.

b. Activation Parameters for C-Cl Reductive Eliminationfrom 1. Transition state B has been located at ∆G q298 ) 21.2kcal ·mol-1 (∆H q ) 18.9 kcal ·mol-1), higher in energy than

(29) (a) Petrie, S.; Stranger, R. Inorg. Chem. 2004, 43, 2597–2610. (b)Cramer, C. J.; Truhlar, D. G. Phys. Chem. Chem. Phys. 2009, 11,10757–10816.

(30) Zhao, Y.; Truhlar, D. G. Org. Lett. 2007, 9, 1967–1970.(31) (a) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215–241.

(b) Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157–167.(32) (a) Benitez, D.; Tkatchouk, E.; Goddard, W. A., III Organometallics

2009, 28, 2643–2645. (b) Benitez, D.; Tkatchouk, E.; Gonzalez, A. Z.;Goddard, W. A., III; Toste, F. D. Org. Lett. 2009, 11, 4798–4801. (c)Benitez, D.; Shapiro, N. D.; Tkatchouk, E.; Wang, Y. M.; Goddard,W. A., III; Toste, F. D. Nat. Chem. 2009, 1, 482–486.

(33) Benitez, D.; Tkatchouk, E.; Yoon, I.; Stoddart, J. F.; Goddard, W. A.,III J. Am. Chem. Soc. 2008, 130, 14928–14929.

Scheme 5. Oxidation of 23 with Unsymmetrical Peroxide 30Affords a 2:1 Mixture of 26 and 27, which Establishes the KineticPreference for Reductive Elimination of More Electron-RichBenzoate Ligands

Figure 4. First-order rate constant of C-Cl reductive elimination as afunction of [AcOH].



structure A (Figure 5). This agrees well with the experimentallydetermined activation parameters (∆G q298 ) 20.5 ( 0.1kcal ·mol-1 and ∆H q) 17.2 ( 2.7 kcal ·mol-1, Table 1).

Unlike in structure A, the Pd nuclei in B are non-equivalentbecause C-Cl bond formation proceeds at Pda while ionizationof Cl- proceeds at Pdb. The Pd-Pd distance is longer in B (2.65Å) than in A (2.62 Å) by 0.03 Å. The bond lengths betweenPda and the carbon (Ca) and chlorine (Cla) ligands that participatein reductive elimination are longer by 0.20 and 0.12 Å in thetransition state, respectively. The Pd-O bond trans to Ca is 0.16Å longer in B than it is in A.

Structure C has been located as a local energy minimum inthe gas phase; C is not an energy minimum when a CH2Cl2

solvent model is applied but is depicted in Figure 5 to betterillustrate the path of reductive elimination. In C, the C-Cl bondis 1.76 Å long, indicating that bond formation is nearly complete(for comparison, the C-Cl bond in E is 1.75 Å). Ionization ofClb from C affords ion pair D, which is 6.8 kcal ·mol-1 higherin energy than A. Experimentally, the structure of the Pd(II)complex immediately following reductive elimination has notbeen determined. Computationally, the ion pair D evolves toPd(II) structure E. Several other Pd(II) complexes could formby ligand rearrangement, and structure E was only chosen as astationary point in Figure 5 to illustrate that, as observed by

experiment, reductive elimination from a dinuclear Pd(III)complex to form a Pd(II) complex, such as E, is energeticallyfavorable.

c. Comparison of Computed Transition State with HammettAnalyses. We have computed natural charge distributions ofstructure A and transition state B (Figure 6) from naturalpopulation analyses.34 The difference in computed chargedistribution in the transition state B and structure A wascompared with the experimentally determined F-values fromthe Hammett analyses. The relative charge distributions in Aand B are in qualitative agreement with the experimentallydetermined F-values. For example, the increased anionic chargein the transition state of the Ca atom bound to Pda (0.05 for A;-0.04 for B) is consistent with the positive F-value (+1.46)that was experimentally determined for benzo[h]quinolinylligand substitution.

Discussion

In the following discussion we will first analyze experimentaland computational evidence consistent with reductive elimina-tion proceeding from a dinuclear palladium complex. Subse-quently, we will evaluate the redox participation of both metalsduring reductive elimination (metallicity). Because metallicitycannot be probed experimentally, we have pursued a compu-tational description of metallicity.

Nuclearity of Pd Complex during Redox Transformation.C-Cl bond formation from complex 1 could proceed viareductive elimination from a mono- (after dissociation of 1) ora dinuclear complex (Figure 7).

Reductive elimination from a mononuclear complex followinginitial Pd-Pd bond cleavage could, in principle, afford observed

Figure 5. Calculated gas-phase stationary points in the mechanism of C-Cl reductive elimination from A. Energies are solvent-corrected electronic energiesat 0 K (E0 K); ∆H q and ∆G q (transition state B) for the reductive elimination are at 298 K.

Table 1. Comparison of Experimental and ComputationalActivation Parameters for Reductive Elimination from 1

∆H q (kcal · mol-1) ∆S q (cal · K-1) ∆G q298 (kcal · mol-1)

experiment 17.2 ( 2.7 -11.2 ( 9.4 20.5 ( 0.1computation 18.9 -7.5 21.2

Figure 6. Partial charge distribution in structure A and transition state B(numbers in parentheses). Hammett analyses of benzo[h]quinolinyl, bridgingcarboxylate, and apical ligands are consistent with the computed chargedistributions.

Figure 7. C-Cl reductive elimination could, in principle, proceed frommononuclear or dinuclear complexes.



product 2.35 We have considered two processes by which thedinuclear core could fragment into mononuclear palladiumcomplexes prior to reductive elimination: (1) homolytic Pd-Pdbond cleavage to afford two mononuclear Pd(III) complexesand (2) heterolytic Pd-Pd bond cleavage to afford one Pd(IV)complex and one Pd(II) complex (Figure 8). For each pathway,either Pd-Pd bond cleavage or subsequent reductive eliminationfrom mononuclear Pd complexes could be rate determining.

Fast homolytic Pd-Pd bond cleavage to afford two identicalPd(III) monomers, followed by rate-determining reductiveelimination, would proceed with a reaction order of 0.5 withrespect to 1 and was excluded by the observed first-order ratelaw for decomposition of 1 and formation of 2.18

Rate-determining homolytic Pd-Pd bond cleavage, followedby fast reductive elimination, was excluded by comparison ofthe activation parameters for C-Cl reductive elimination (∆H q

) 17.2 ( 2.7 kcal ·mol-1, ∆S q ) -11.2 ( 9.4 cal K-1, and∆G q298 ) 20.5 ( 0.1 kcal ·mol-1) with the calculated Pd-Pdbond strength. The Pd-Pd bond in A was calculated to be 35.3kcal ·mol-1 by calculating the energy required to separate thepalladium nuclei to 3.65 Å. Stretching of the Pd-Pd bond ledto geometrical distortions of the bridging acetate ligands.Therefore, the Pd-Pd bond energy was also calculated forstructure F, in which the bridging acetate ligands have beenreplaced by chelating propanedialato ligands (33.2 kcal ·mol-1)(eq 4). The computed value of 35.3 kcal ·mol-1 is likely anunderestimate of the energy required to cleave the dinuclearcore because cleavage requires breaking of the Pd-Pd bond inaddition to two Pd-O bonds. The effect of solvent polarization(Poisson-Boltzmann model) in CH2Cl2 was included whilecalculating the Pd-Pd bond strength in F to account for solventstabilization of the complexes generated after bond cleavage.

Reductive elimination could potentially proceed by dispro-portionation of 1 into a Pd(II) and a Pd(IV) complex, withsubsequent monometallic reductive elimination from a mono-

nuclear Pd(IV) complex (Path 2, Figure 8). Such dispropor-tionative metal-metal bond cleavage has been reported fordinuclear Pt(III) intermediates during the oxidation of Pt(II) toPt(IV).36 As with homolytic Pd-Pd bond cleavage, eitherheterolytic cleavage or subsequent reductive elimination couldbe rate limiting.

Possible pre-equilibrium dimer cleavage, followed by rate-determining reductive elimination, was evaluated by a crossoverexperiment between acetate-bridged complex 1 and benzoate-bridged complex 20a (Scheme 6). Complex 33, the product ofcrossover between 1 and 20a, was not observed during C-Clreductive elimination, precluding pre-equilibrium dimer cleav-age. Addition of 1.0 equiv of benzo[h]quinoline (8), a potentiallycoordinating ligand, did not lead to observed exchange either.

To evaluate the possibility of rate-determining heterolyticPd-Pd cleavage prior to C-Cl reductive elimination, therelative rates of reductive elimination from acetate-bridgedcomplex 1 and esp-bridged complex 35 (eq 5) were measuredat 35 °C (R,R,R′,R′-tetramethyl-1,3-dibenzenedipropionate)(esp). Reductive elimination from esp-bridged Pd(III) complex35 proceeds 16 times more slowly than that from acetate-bridgedPd(III) complex 1. However, the esp-bridged Pd(II) complex34 dissociates at least 1500 times more slowly than acetate-bridged Pd(II) complex 9 (see Supporting Information). Therelative rates of reductive elimination from 1 and 35 aretherefore most consistent with reductive elimination proceedingfrom a dinuclear complex. In addition to the experimental data,we computed the barrier for heterolytic cleavage of the dinuclearcore to generate a Pd(II) and Pd(IV) complex to be at least 29.5kcal ·mol-1 (see Supporting Information); the experimentallydetermined activation enthalpy for reductive elimination from1 is H q ) 17.2 ( 2.7 kcal ·mol-1.

The combination of the observed first-order disappearanceof 1 and formation of 2, the measured activation parameterswhich show that reductive elimination from 1 is more facilethan homolytic Pd-Pd bond cleavage, and the lack of crossoverbetween 1 and 20a are all consistent with reductive eliminationfrom a dinuclear complex. Further, the computed reactionpathway agrees with experimental observations of ground-statestructure, Arrhenius parameters for C-Cl reductive elimination,

(34) (a) Landis, C. R.; Weinhold, F. J. Comput. Chem. 2007, 28, 198–203. (b) Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys.1985, 83, 735–746.

(35) Schore, N. E.; Ilenda, C.; Bergman, R. G. J. Am. Chem. Soc. 1976,98, 7436–7438.

(36) (a) Bandoli, G.; Caputo, P. A.; Intini, F. P.; Sivo, M. F.; Natile, G.J. Am. Chem. Soc. 1997, 119, 10370–10376. (b) Canty, A. J.; Gardiner,M. G.; Jones, R. C.; Rodemann, T.; Sharma, M. J. Am. Chem. Soc.2009, 131, 7236–7237.

Figure 8. Potential pathways via mononuclear Pd complexes. Path 1:Dissociation into two Pd(III) monomers followed by reductive eliminationfrom mononuclear Pd(III). Path 2: Disproportionation to Pd(II) and Pd(IV)followed by reductive elimination from Pd(IV).

Scheme 6. Exchange between 1 and 20a is Not Observed on theTime Scale of Formation of 2, Precluding Pre-equilibriumDissociation of 1 during Reductive Elimination



and Hammett F-values and also implicates dinuclear complexesin every step of reductive elimination.

Metallicity of Reductive Elimination. Knowledge of thenuclearity of the complex from which reductive eliminationproceeds provides no information about the role of each metalcenter during reductive elimination. Depending on the extentof metal-metal cooperation during reductive elimination froma dinuclear complex, either monometallic or bimetallic redoxpathways are possible (Figure 9).

To probe the electronic contribution of each metal centerduring reductive elimination, the electron binding energies37 ofthe 4s core electrons of both Pda and Pdb were calculated as afunction of reaction progress for pathways 3-5 shown in Figure9. The electron binding energy38 of a 4s electron is the energyrequired to remove an electron from the 4s orbital42 to infiniteseparation, without allowing any relaxation of the remainingelectrons. For metal centers with similar ligand environments,electron binding energy is well correlated with formal oxidationstate.43 The electron binding energy decreases during reductionof a metal center because the metal becomes more electron richand the electron is more effectively shielded from the nuclear

charge. Conversely, the electron binding energy increases duringoxidation because the electron is less effectively shielded fromthe nuclear charge. When comparing the electron bindingenergies in metal centers with similar ligand sets, higher electronbinding energies correlate with higher oxidation states. Wesuggest that the electron binding energies, which correlate withoxidation state, are suitable to ascertain the redox participationof the metals.44

We selected two hypothetical but well-defined limiting casesfor monometallic reductive elimination (paths 3 and 5, Figure9) and computed the pathways of reductive elimination for both.Subsequently, the electron binding energies of electrons in the4s orbitals of both palladium centers, Pda and Pdb, have beencalculated for each of the two pathways.45

Hypothetical reductive elimination from Pda with Pdb as anon-redox-active spectator ligand on Pda would proceed througha dinuclear Pd(I)/Pd(III) mixed-valence complex (Figure 10a,path 3). We investigated path 3 by computationally cleaving Ainto two mononuclear Pd(III) complexes (G). Computedintermediate G serves as model for dinuclear compound A, inwhich potential metal-metal cooperation is eliminated. Theelectron binding energies were determined computationally todescribe the redox chemistry of each Pd center during hypo-thetical path 3 (Figure 10a). The cleavage from A to G is redox-neutral and does not significantly affect the computed electronbinding energy of either Pda or Pdb. Monometallic reductiveelimination from Pda proceeds through transition state H toafford structure I. Redox chemistry is occurring only at Pda

during the transformation of G to I, and thus the electron bindingenergy of Pdb does not change. At the same time, the electronbinding energy of Pda decreases as G is transformed to I,consistent with redox contribution at Pda. We selected the Pd(II)complex J as the end point, because complex J has well-definedoxidation states on palladium. Further, the ligand sphere ofpalladium in J is closely related to the ligand sphere of palladiumin A, allowing for meaningful comparison of electron bindingenergies. Formal comproportionation of the mixed-valencestructure I to J results in the electron binding energies of Pda

and Pdb converging. Pda in structure I is most accuratelydescribed as a Pd(II) with a ligand-centered radical, and thusthe electron binding energy of Pda during comproportionationdoes not increase, as would be expected for oxidation of Pd(I)to Pd(II). The diagram in Figure 10a shows that only Pda(37) The core and valence (available in the LACV3P++**(2f) basis set)

natural atomic orbital energies of both Pda and Pdb were obtained andplotted as a function of reaction progress. According to Janak’stheorem, the relative orbital energies are estimates of the relativevertical electron binding energies. The relative natural atomic orbitalenergies can be regarded as approximations of the vertical ionizationpotential or the negative of the electron affinity.

(38) Energies were obtained from Natural Bond Orbital (NBO)39analysesof the M06 wavefunction using the NBO 5.0 module40 in Jaguar. TheNBO analysis method is a popular mathematical treatment of wave-functions to study bonding effects such as hybridization andcovalency.32c,41

(39) (a) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. ReV. 1988, 88,899–926. (b) Weinhold, F.; Landis, C. R. Valency and Bonding: ANatural Bond Orbital Donor-Acceptor PerspectiVe; Cambridge Uni-versity Press: Cambridge, UK, 2005.

(40) Glendening, E. D.; Badenhoop, J. K.; Reed, A. E.; Carpenter, J. E.;Bohmann, J. A.; Morales, C. M.; Weinhold, F. Jaguar; TheoreticalChemistry Institute, University of Wisconsin: Madison, 2001.

(41) (a) Kim, C. K.; Lee, K. A.; Kim, C. K.; Lee, B.-S.; Lee, H. W. Chem.Phys. Lett. 2004, 391, 321–324. (b) Nori-Shargh, D.; Roohi, F.;Deyhimi, F.; Naeem-Abyaneh, R. J. Mol. Struct. 2006, 763, 21–28.(c) Leyssens, T.; Peeters, D.; Orpen, A. G.; Harvey, J. N. Organo-metallics 2007, 26, 2637–2645.

(42) The 4s orbital was selected because it is spherically symmetrical andmore sensitive to perturbations in the electron density of the metalcenter than energetically lower lying s orbitals.

(43) (a) Riggs, W. M. Anal. Chem. 1972, 44, 830–832. (b) Leigh, G. J.Inorg. Chim. Acta 1975, 14, L35–L36. (c) Chatt, J.; Elson, C. M.;Hooper, N. E.; Leigh, G. J. J. Chem. Soc., Dalton Trans. 1975, 2392–2401. (d) Cisar, A.; Corbett, J. D.; Daake, R. L. Inorg. Chem. 1979,18, 836–843. (e) Corbett, J. D. Inorg. Chem. 1983, 22, 2669–2672.

(44) Similarly analysis of the carbon 1s orbitals of CH3 bound to transitionmetals has been shown to correlate well with the electrophilicity ornucleophilicity of the methyl groups. Nielsen, R. J.; Goddard, W. A.,III. J. Phys. Chem., manuscript in preparation.

(45) In addition to calculating the electron binding energy from A, we havealso computed the electron binding energies of both Pd centers duringC-Cl reductive elimination from cationic structure M, in which oneof the apical chloride ligands is replaced by a pyridyl ligand. Theresults of these calculations are in the Supporting Information, andthe conclusions drawn from this structure mirror those from complexA.

Figure 9. Potential pathways from dinuclear Pd complexes. Path 3:Monometallic reductive elimination to form a Pd(I)/Pd(III) mixed-valencecomplex and 2, followed by comproportionation to afford Pd(II). Path 4:Computed low-energy pathway for reductive elimination (Figure 5). Path5: Disproportionation to a Pd(II)/Pd(IV) mixed-valence complex, followedby monometallic reductive elimination.



participates in redox chemistry during reductive elimination;hypothetical path 3 is therefore monometallic.

Alternatively, C-Cl bond formation could proceed by initialdisproportionation of complex 1 to a mixed-valence Pd(II)/Pd(IV) complex prior to reductive elimination from Pd(IV), asshown in Figure 9, path 5. Previously, it was speculated thatreductive elimination from dinuclear Pd(III) complexes mayproceed via Pd(II)/Pd(IV) mixed-valence species.46,47 As shownin Figure 10b, the electron binding energy diagram computedfor disproportionation from A to K shows increased electronbinding energy of Pda because Pda is more oxidized in K thanit is in A. Reductive elimination proceeds through transitionstate L, in which redox chemistry is occurring at Pda. Theelectron binding energies of Pda and Pdb converge as reductiveelimination affords two Pd(II) centers (J). During reductiveelimination from Pda, the electron binding energy of Pdb is

constant because Pdb is not redox-active according to path 5.Hypothetical path 5 is therefore monometallic.

Both paths 3 and 5 were selected to illustrate and benchmarkthe electron binding energies in two limiting cases of mono-metallic reductive elimination. Path 4 is derived from thecomputed low-energy pathway shown in Figure 5. Both electron

(46) Deprez, N. R.; Sanford, M. S. J. Am. Chem. Soc. 2009, 131, 11234–11241.

(47) The relative energy of acetate-bridged Pd(II) dimers in which eitherthe Pd centers are held in proximity (as in 9) or the palladium centersare in a planar arrangement (as in K) was investigated: Jack, T. R.;Powell, J. Can. J. Chem. 1975, 53, 2558–2574. It was concluded thata planar complex is greater than 12 kcal ·mol-1 higher in energy thana structure in which the palladium centers are held in proximity. Wehave been unable to locate a local minimum for a planar complexwith two Pd(II) centers; all attempts to calculate such a species resultin the structure folding to bridged structure 9. Structure K is calculatedto be 12.4 kcal ·mol-1 higher in energy than A.

Figure 10. (a) Plot of the 4s core electron binding energy of Pda (solid) and Pdb (dashed) as a function of reaction progress for monometallic reductiveelimination via a Pd(I)/Pd(III) mixed-valence species (Figure 9, path 3). (b) Plot of electron binding energy of Pda and Pdb as a function of reaction progressfor monometallic reductive elimination from a Pd(II)/Pd(IV) mixed-valence species (Figure 9, path 5). (c) Plot of electron binding energy of Pda and Pdb asa function of reaction progress for calculated bimetallic reductive elimination (Figure 5).



binding energies of Pda and Pdb monotonically decrease whileA is converted to C via transition state B. The electron bindingenergies continue to monotonically decrease until Pd(II) struc-ture J. Neither metal center becomes oxidized beyond Pd(III)during the course of reductive elimination.

The calculated electron binding energies implicate simulta-neous redox participation of both metals during reductiveelimination. Monotonic reduction in electron binding energy atboth metals is inconsistent with initial disproportionation to aPd(II)/Pd(IV) complex prior to monometallic reductive elimina-tion. A simultaneous change in the electron binding energiesof both Pda and Pdb is also inconsistent with initial monometallicreductive elimination to a Pd(I)/Pd(III) mixed-valence complex.Therefore, we conclude that reductive elimination from 1 isbimetallic.

Kinetic Advantage of Bimetallic Reductive Elimination.Based on the HOMO and LUMO calculated for structure Aand transition state B (see Supporting Information), metal-metalbonding during reductive elimination is accomplished primarilyby overlap of the palladium dz2 orbitals. We have calculatedthe activation barrier for reductive elimination as a function ofPd-Pd distance. Computationally forced elongation of thePd-Pd bond attenuates the overlap of the dz2 orbitals and thusdecreases electronic communication between the metals ascompared to low-energy structure A (eq 6). By preventingelectronic communication between the metals, bimetallic reduc-tive elimination is forced to become increasingly monometallic.

The activation barrier for reductive elimination when thePd-Pd distance is fixed at 2.62 Å (the Pd-Pd distance in A) is17.1 kcal ·mol-1. Computational elongation of the Pd-Pddistance to 2.95 Å (Table 2, entry 2) increased the activationbarrier to 21.8 kcal ·mol-1. Further elongation to 3.30 and 3.65Å led to activation barriers of 33.3 and 46.0 kcal ·mol-1,respectively (entries 3 and 4). Pd-Pd distances beyond 3.65Å, which is greater than the sum of the two van der Waals radiiof the palladium atoms, could not be accommodated whilemaintaining the bridging geometry of the acetate ligands.Comparison of the activation barriers for bimetallic reductiveelimination (entry 1) and monometallic reductive elimination(entry 4) indicates that metal-metal cooperation during reductiveelimination lowers the energy barrier by ∼30 kcal ·mol-1 ascompared to monometallic reductive elimination, which ap-proximately corresponds to the Pd-Pd bond dissociation energyin 1.

Reductive Elimination in the Presence of AcOH. Previously,on the basis of the observation of C-C,48 C-O,2d,e C-Cl,49

and C-F50 reductive elimination from isolated Pd(IV) com-plexes, Pd(II)/Pd(IV) redox cycles have been invoked asmechanistic rationales for Pd-catalyzed C-H oxidation reac-tions.51 We have proposed that reductive elimination fromdinuclear Pd(III) complexes, similar to 1, is the product-formingstep in a variety of Pd-catalyzed C-H oxidation reactions onthe basis of the observation of dinuclear complexes duringchlorination3a and acetoxylation3b of 2-phenylpyridine deriva-tives.52

During catalysis, C-H metalation by Pd(II) generates anequivalent of acid. We have found that the rate of reductiveelimination is linearly correlated with [AcOH]. We propose thatthe observed C-Cl reductive elimination rate enhancement isdue to protonation of one of the bridging acetate ligands of 1to afford 36 (Figure 11). In the presence of acid, protonation ofthe acetate ligand would result in a pentacoordinate Pd centerof a dinuclear complex, in which one Pd-O bond is cleaved.We suggest that reductive elimination from 36, in which onebridging acetate ligand is protonated, is faster than that from 1.A dinuclear transition state for reductive elimination fromcationic complex 36, similar in structure to the transition statefor reductive elimination from 1 (B), has been located compu-tationally. The barrier to reductive elimination from 1 wascomputed to be 21.8 kcal ·mol-1, whereas the barrier to reductiveelimination from 36 was computed to be 20.1 kcal ·mol-1,consistent with the observed rate enhancement in the presenceof AcOH.45 Computed electron binding energies for bothpalladium atoms in A + H+ and in B + H+ indicate that, similarto reductive elimination from A, reductive elimination from aprotonated dinuclear core is bimetallic as well. During Pd-catalyzed C-H functionalization, C-H metalation generatesacid, and thus acid-catalyzed reductive elimination may be

(48) Byers, P. K.; Canty, A. J.; Skelton, B. W.; White, A. H. J. Chem.Soc., Chem. Commun. 1986, 1722–1724.

(49) Whitfield, S. R.; Sanford, M. S. J. Am. Chem. Soc. 2007, 129, 15142–15143.

(50) (a) Furuya, T.; Ritter, T. J. Am. Chem. Soc. 2008, 130, 10060–10061.(b) Furuya, T.; Benitez, D.; Tkatchouk, E.; Strom, A. E.; Tang, P.;Goddard, W. A., III; Ritter, T. J. Am. Chem. Soc. 2010, 132, 3793–3807.

(51) (a) Giri, R.; Chen, X.; Yu, J.-Q. Angew. Chem., Int. Ed. 2005, 44,2112–2115. (b) Jordan-Hore, J. A.; Johansson, C. C. C.; Gulias, M.;Beck, E. M.; Gaunt, M. J. J. Am. Chem. Soc. 2008, 130, 16184–16186.(c) Qin, C.; Lu, W. J. Org. Chem. 2008, 73, 7424–7427. (d) Neumann,J. J.; Rakshit, S.; Droge, T.; Glorius, F. Angew. Chem., Int. Ed. 2009,48, 6892–6895. (e) Wang, X.; Mei, T.-S.; Yu, J.-Q. J. Am. Chem.Soc. 2009, 131, 7520–7521. (f) Zhang, Y.-H.; Shi, B.-F.; Yu, J.-Q.Angew. Chem., Int. Ed. 2009, 48, 6097–6100. (g) Wang, G.-W.; Yuan,T.-T. J. Org. Chem. 2010, 75, 476–479.

(52) A detailed investigation of the relevance of dinuclear Pd(III) complexesin catalysis will be published separately: Powers, D. C.; Xiao, D. Y.;Geibel, M. A. L.; Ritter, T. J. Am. Chem. Soc. manuscript accepted.

Figure 11. Reductive elimination from 1 in the presence of AcOH proceedsvia at least two pathways, one linearly dependent on and one independentof [AcOH].

Table 2. Computed Reference Ground-State and ActivationEnergies of Reductive Elimination as a Function of RestrictedPd-Pd Distancea

Pd-Pd distance (Å)ground-state

energy (kcal · mol-1)activation energy

(kcal · mol-1) relative to A

2.62 0.0 (A) 17.12.95 6.2 21.83.30 20.5 33.33.65 35.3 46.0

a Energies are gas-phase electronic energies at 0 K (E0 K).



relevant during catalysis. On the basis of the presented data,we cannot distinguish between pre-equilibrium and rate-determining protonation in the acid-catalyzed pathway forreductive elimination from 1 (k′, Figure 11).

Conclusions

In this article we sought to address the following questions:Does the core of dinuclear Pd(III) complex 1 stay intact duringreductive elimination? Do both metal centers participate in theredox transformation, and is there an energetic benefit to havingtwo metals as opposed to a single metal? How closely relatedare previously proposed mononuclear Pd(IV) complexes to thedinuclear Pd(III) complexes reported here with respect to themechanism of reductive elimination?

Experimental and theoretical analysis of C-Cl reductiveelimination from complex 1 has implicated reductive eliminationfrom a dinuclear core with synergistic, bimetallic redoxparticipation of both metals during reductive elimination. Thepresence of the second metal in dinuclear complex 1 lowersthe activation barrier of reductive elimination. If reductiveelimination is artificially forced to proceed via a monometallicpathway by eliminating metal-metal communication, the barrierto reductive elimination is increased by ∼30 kcal ·mol-1.

Transition-metal-mediated C-heteroatom bond constructionremains a synthetically challenging goal. We have shown that

C-heteroatom bonds can be efficiently formed by bimetallicreductive elimination from dinuclear Pd(III) complexes. In thisarticle we have provided a theoretical framework for bimetallicredox catalysis. Redox synergy between metal centers, as is seenfor C-Cl reductive elimination here, also has the potential tolower the activation barriers for other redox processes. Weanticipate that our findings may serve as a foundation for futurereaction development.

Acknowledgment. We thank Douglas M. Ho, Jessica Y. Wu,and Takeru Furuya for X-ray crystallographic analysis, RobertNielsen and David Ford for insightful discussions, Sanofi Aventisfor a graduate fellowship for D.C.P., the reviewers of thismanuscript for comments, which improved the quality of themanuscript, and the NSF (CHE-0952753) and NIH-NIGMS(GM088237) for funding. Computational facilities were funded bygrants from ARO-DURIP and ONR-DURIP.

Supporting Information Available: Detailed experimentalprocedures and spectroscopic data for all new compounds;detailed computational methods and XYZ coordinates (PDF,CIF). This material is available free of charge via the Internetat http://pubs.acs.org.

JA1036644



Documents

Bimetallic Reductive Elimination from Dinuclear Pd(III) Complexeswag.caltech.edu/publications/sup/pdf/886.pdf · 2010-11-02 · Bimetallic Reductive Elimination from Dinuclear Pd(III)